Every persona. Every step.
Every decision.
The engine that runs each AI persona through your product flow step by step — tracking frustration, trust, and attention in real time — deciding exactly when and why they leave.
How the Engine is Built
Behavioral State Machine
Pure code. No AI. Tracks each persona's state: frustration score, attention budget, trust level, steps taken. When any threshold is crossed, the persona drops off. Fast, deterministic, testable.
LLM as UX Interpreter
The AI reads your product's UI — a form, a button label, a pricing page — and translates it into a behavioral event. "8-field form with no progress bar" becomes a cognitive overload event. The state machine acts on it.
Live Persona State
"Payment screen asked for Aadhaar before showing total amount. No Hindi option. OTP step visible."
What Makes Behavior Realistic
Frustration compounds
Each friction point adds to a running score. No single moment kills a session — the threshold does.
Attention decays
Personas start with a fixed attention budget. Every slow screen and long form burns it down. When it hits zero, they leave.
Trust can collapse
Certain triggers — hidden charges, Aadhaar requests, no social proof — cause sudden trust drops, not gradual ones.
Noise is built in
15–20% of decisions go against the model. Real users are irrational. Our personas are too.
"Most simulation tools model the ideal user. We model the real one — impatient, suspicious, and easily confused."
Ready to test your product? Head to the validation page to run your first simulation.